The new Politico/GWU/Battleground poll seems to me, from a quick perusal of its internals, to have produced solid and non-quirky results consistent with several other surveys. It has a D+3 sample, and shows an Obama margin of 3 on the presidential ballot test and a 1 point Democratic edge on the generic congressional ballot.
Which raises the question: Is the conventional assumption that Republicans will continue to hold the House sound?
It may not be. Two other recent likely voter polls have produced an R+1 and a tied generic congressional ballot. So let’s say that right now the congressional ballot is tied. The closest we’ve come to an even national popular vote for the House in recent years was in 2000, when Republicans had a narrow popular vote margin of .3 percent, and ended up with a narrow 221-212 margin in seats. An even popular vote tends to translate into pretty even results in seats split between the two parties. In the wave elections of 2006, 2008, and 2010, by contrast, the parties’ popular vote margins ranged from 6 to 8 percentage points. The middling GOP majorities of 2002 and 2004 were based on national popular vote margins of more than 2 1/2 points.